About MolTrust

// Definition

What is MolTrust?

MolTrust is a trust infrastructure platform for AI agents. It issues W3C Verifiable Credentials to AI agents via a Model Context Protocol (MCP) Server, anchors every credential cryptographically on the public ledger, and exposes REST API endpoints that allow agents and services to verify agent identity in real time. MolTrust is operated by CryptoKRI GmbH, Zürich, Switzerland.

AI agents are increasingly autonomous — they transact, delegate, and make decisions without human oversight at every step. No standardized trust layer exists to answer the basic question: who is this agent, and can it be trusted? MolTrust answers that question with cryptographic certainty, using open standards that any system can verify independently.

// Technical Standards

What standards does MolTrust implement?

MolTrust is built exclusively on open, ratified standards. Credentials are verifiable by any W3C-compliant resolver — no vendor lock-in, ever.

W3C DID v1.0

Decentralized Identifiers — self-sovereign, cryptographically verifiable agent identity. did:moltrust DID Method Spec

W3C VC Data Model v2.0

Verifiable Credentials in JSON-LD format — tamper-evident, issuer-signed attestations about agent capabilities.

Model Context Protocol

36 tools across 6 verticals, callable by any MCP-compatible AI agent or client.

x402 Protocol

HTTP 402-based micropayment protocol on a public ledger. Agents pay $1–$5 USDC per credential issuance — no subscription required.

ERC-8004

Agent identity standard for EVM chains — MolTrust credentials are ERC-8004 compatible.

Base Blockchain

Ethereum L2 by Coinbase. Every issuance is anchored with a SHA-256 hash and transaction receipt on Base mainnet.

Singapore IMDA MGF

Aligned with the Singapore IMDA Model AI Governance Framework for Agentic AI (January 2026) — the first governance standard for autonomous AI agents.

Protocol Compliance v0.2.2

ViolationRecord registry, delegation chain depth-limit (8 hops), sequential signing validation — March 2026.

// How it works

How does MolTrust issue credentials?

The full issuance flow — from API call to on-chain anchor — completes in under 3 seconds.

Agent calls the API

An AI agent sends a request to a MolTrust endpoint, e.g. POST /vc/skill/issue, with its DID and the capability to certify.

8-point security audit

MolTrust runs 8 automated checks — endpoint responsiveness, rate limiting, auth handling, error leakage, DID resolution, schema compliance, and more.

x402 micropayment

The agent pays $1–$5 USDC via the x402 protocol on a public ledger. No pre-registration, no subscription, no API key required.

Credential issuance & immutable audit anchoring

MolTrust issues a signed W3C Verifiable Credential, anchors a SHA-256 hash on Base mainnet, and returns the credential with a BaseScan transaction link.

Any party can verify — without MolTrust

Verifiers call GET /skill/verify/:hash or resolve the DID with any W3C-compliant resolver. No dependency on MolTrust infrastructure.

// Protocol Architecture

Five layers of agent trust

MolTrust implements a complete trust stack. Each layer builds on the one below — from identity to non-repudiation.

Non-Repudiation NEW

Interaction Proof Records (IPR) — Merkle-batched, on-chain anchored. Proves what an agent did, when, with cryptographic certainty. No party can deny the interaction occurred.

Reputation

Swarm Intelligence Protocol — peer endorsements, cross-vertical trust propagation, prediction accuracy tracking. Decentralized scoring without a central authority.

Authorization

W3C Verifiable Credentials — BuyerAgent, TravelAgent, SkillVerification, Salesguard. Ed25519-signed, 7-day expiry, delegation chains up to 8 hops.

L2.5

Sequential Safety NEW

Sequential Action Safety (SAS) — deterministic pre-execution check for order-sensitive action sequences. Detects when individually permitted actions produce catastrophic outcomes in the wrong order. Opt-in, Phase 1 WARN-only.

Identity

W3C DID:web + ERC-8004 on-chain registration. Self-sovereign agent identity anchored on Base L2. Resolvable by any W3C-compliant DID resolver.

Layer 4 — How IPR works

Every agent interaction produces an IPR: a SHA-256 hash of the output + agent DID + timestamp + confidence score. IPRs are batched into a Merkle tree and anchored as a single Base L2 transaction — cost-efficient and independently verifiable.

APS Agent Receipt

An aeoess agent processes a transaction. The output hash is submitted as an IPR via POST /vc/ipr/submit. The receipt includes the Merkle proof for independent verification.

VCOne Verification Receipt

VCOne verifies a credential chain. The verification result + delegation path are hashed and submitted as an IPR. Any auditor can verify VCOne's action occurred at the claimed time.

MolTrust Credential Anchor

When MolTrust issues a VC, the credential hash becomes an IPR. The Merkle batch transaction proves issuance time, preventing backdating or credential tampering.

—

IPRs recorded

—

On-chain anchored

—

Agents tracked

Base L2

Anchor chain

// Solution Portfolio

Which verticals does MolTrust support?

Each vertical ships with its own Verifiable Credential schema, REST endpoints, and MCP tools. The table below updates automatically when new verticals are added to the PORTFOLIO config.

Vertical	Credential Type	Key Capability	Page

// FAQ

Frequently Asked Questions

MolTrust is for developers building AI agents that need verifiable identity, for platforms that need to verify agent credentials before granting access, and for enterprises deploying autonomous agent workflows that require auditable trust chains. MolTrust provides an MCP Server, REST API, and npm middleware package.

MolTrust anchors all credentials on Base mainnet — an Ethereum L2 operated by Coinbase. Payments are processed in USDC via the x402 protocol. MolTrust does not hold customer funds and is not a CASP (Crypto Asset Service Provider) under EU MiCA regulation.

The MolTrust MCP Server is available on GitHub under MoltyCel/moltrust. Credential schemas are published as open JSON-LD specifications. The core API and issuing infrastructure is proprietary and operated by CryptoKRI GmbH, Zürich.

Several endpoints are free: /skill/audit, /agent/score-free/:addr, /health, and all schema endpoints. Paid endpoints cost $1–$5 USDC per call via x402. No subscription or account required — agents pay per use.

MolTrust is operated by CryptoKRI GmbH, registered in Zürich, Switzerland. The platform is accessible globally at moltrust.ch and api.moltrust.ch.

Yes. MolTrust credentials are W3C Verifiable Credentials resolvable by any standards-compliant DID resolver. The on-chain anchor on Base mainnet means anyone can independently verify authenticity and issuance timestamp — no dependency on MolTrust infrastructure.